You can administer NCP from the NSX Manager GUI or from the command-line interface (CLI).

Note: If a container host VM is running on ESXi 6.5 and the VM is migrated through vMotion to another ESXi 6.5 host, containers running on the container host will lose connectivity to containers running on other container hosts. You can resolve the problem by disconnecting and connecting the vNIC of the container host. This issue does not occur with ESXi 6.5 Update 1 or later.

Hyperbus reserves VLAN ID 4094 on the hypervisor for PVLAN configuration and this ID cannot be changed. To avoid any VLAN conflict, do not configure VLAN logical switches or VTEP vmknics with the same VLAN ID.

If the nsx-ovs container running in the nsx-node-agent pod is restarted for any reason, Kubernetes services might become unavailable for 2 minutes or more. This is expected behavior.

Do not recreate a pod with the same pod name if the NCP pod is stopped and the node agent pod is running. The new pod will have the wrong network configuration and its network traffic will fail.